What it does
Meilisearch indexes documents and ranks results using configurable search rules. It supports facets, filters, and hybrid search features.
Why it’s interesting
Ranking and indexing choices determine how search feels. A small local dataset is enough to explore typo tolerance and filtering.
Real-world use case
Index a small product catalog and search it with an intentional typo.
What I can learn
- 01
Inverted indexes and bitmap filtering
See how token-to-document indexes and compact bitmaps narrow the candidate set for text queries and filters.
- 02
Ranking-rule evaluation
Follow the ordered ranking criteria that make typo tolerance, word proximity, and exact matches affect result ordering.
- 03
Hybrid lexical and vector retrieval
Explore how semantic vector matches complement lexical results when users search with related words rather than exact terms.
One thing to explore
Core indexing library
Follow the crates/milli implementation and connect it to the concepts above.
Explore the source (opens in a new tab)10-Minute Challenge
Index and search sample documents
Before you start: Docker; use the current official quickstart for authentication and API-key settings.
- Start the local server using the official quickstart.
- Create an index and add a few sample documents.
- Search with a misspelled query and inspect the matching result.
docker run --rm -p 7700:7700 getmeili/meilisearch:latest